The "load" function could do more of the right thing in the case where an absolute path is given in a load call and then within the file loaded, require or load references are made to relative files whose directories are not in load path but the files do exist within the directory specified by the original load.
So for example, we have a testing directory, "/tmp/" not in load-path with files a.el and b.el.
In a.el, there is the line: (require 'b) and then we do a (load "/tmp/a.el"). The require of b will fail but since load-file-name is set to /tmp/a.el, if no matches are found in load-path, it would be a simple matter to try to resolve b in the directory specified in load-file-name as a fallback.
This would help in many cases when experimenting with new directories of code that have not yet been added to the load path. It also would help the Emacs package system since it adds only the top-level directory of a package to load-path when trying to byte-compile the package. If the package contains subdirectories, which require or load local files within these subdirectories, these will fail right now but would succeed with the above change.
Any comments, suggestions or issues with this? Please consider implementing this and help make multi-directory packages a reality.